bpo-28055: fix unaligned accesses in siphash24() #6123
Conversation
|
Hello, and thanks for your contribution! I'm a bot set up to make sure that the project can legally accept your contribution by verifying you have signed the PSF contributor agreement (CLA). Unfortunately we couldn't find an account corresponding to your GitHub username on bugs.python.org (b.p.o) to verify you have signed the CLA (this might be simply due to a missing "GitHub Name" entry in your b.p.o account settings). This is necessary for legal reasons before we can look at your contribution. Please follow the steps outlined in the CPython devguide to rectify this issue. Thanks again to your contribution and we look forward to looking at it! |
|
A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated. Once you have made the requested changes, please leave a comment on this pull request containing the phrase |
serhiy-storchaka
changed the title
Python/pyhash.c
Outdated
| v3 ^= mi; | ||
| DOUBLE_ROUND(v0,v1,v2,v3); | ||
| v0 ^= mi; | ||
| } | ||
|
|
||
| t = 0; | ||
| pt = (uint8_t *)&t; | ||
| m = (uint8_t *)in; | ||
| m = in; |
There was a problem hiding this comment.
m is not needed. Just use in.
serhiy-storchaka
added
type-bug
|
So other than a NEWS entry, is there anything else that needs to be done for this before merging? |
|
Ping |
| @@ -0,0 +1 @@ | |||
| fix unaligned accesses in siphash24() | |||
There was a problem hiding this comment.
Start the sentence from the title case, add missed period, and please add "Patch by yourname."
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc. https://bugs.gentoo.org/show_bug.cgi?id=636400
|
Thanks @DerDakon for the PR, and @serhiy-storchaka for merging it |
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc. https://bugs.gentoo.org/show_bug.cgi?id=636400 (cherry picked from commit 1e2ec8a) Co-authored-by: Rolf Eike Beer <eike@sf-mail.de>
|
GH-6777 is a backport of this pull request to the 3.7 branch. |
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc. https://bugs.gentoo.org/show_bug.cgi?id=636400 (cherry picked from commit 1e2ec8a) Co-authored-by: Rolf Eike Beer <eike@sf-mail.de>
|
GH-6778 is a backport of this pull request to the 3.6 branch. |
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc. https://bugs.gentoo.org/show_bug.cgi?id=636400 (cherry picked from commit 1e2ec8a) Co-authored-by: Rolf Eike Beer <eike@sf-mail.de>
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc. https://bugs.gentoo.org/show_bug.cgi?id=636400 (cherry picked from commit 1e2ec8a) Co-authored-by: Rolf Eike Beer <eike@sf-mail.de>
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc. https://bugs.gentoo.org/show_bug.cgi?id=636400
Like python#6123 this pointer may be unaligned, so a memcpy() instead of simple assignment is required for strict architectures e.g. sparc.
The hash implementation casts the input pointer to uint64_t* and directly reads from this, which may cause unaligned accesses. Use memcpy() instead so this code will not crash with SIGBUS on sparc.
https://bugs.gentoo.org/show_bug.cgi?id=636400
https://bugs.python.org/issue28055